Is acyl migration to the aglycon avoidable in 2-acyl assisted glycosylation reactions?

Article abstract of DOI:10.1139/v04-059

This report unequivocally separates orthoester formation from acyl transfer for the first time and indicates possible routes to eliminate 2-O-acyl transfer during glycosylation reactions. Experimental evidence is shown that acyl transfer from 2-O-acyl-3,4

Full text of DOI:10.1139/v04-059

1157
Is acyl migration to the aglycon avoidable in 2-
acyl assisted glycosylation reactions?1
Attila Bérces, Dennis M. Whitfield, Tomoo Nukada, Iwona do Santos Z.,
Agnes Obuchowska, and Jiri J. Krepinsky
Abstract: This report unequivocally separates orthoester formation from acyl transfer for the first time and indicates
possible routes to eliminate 2-O-acyl transfer during glycosylation reactions. Experimental evidence is shown that acyl
transfer from 2-O-acyl-3,4,6-tri-O-benzyl-D-galactopyranose-derived glycosyl donors decreases in the order formyl >
acetyl > pivaloyl. The 2-O-benzoyl derivatives are more variable, in some cases transferring easily, and in others not at
all. Density functional theory (DFT) calculations of the structure and energetics of dioxolenium ion and related inter-
mediates suggest that a proton transfer pathway from the nucleophile to O-2 provides an explanation for the observed
trends. These DFT calculations of the proton transfer pathway support a mechanism in which a relay molecule is in-
volved. Further DFT calculations used a constraint based on linear combinations of six bond lengths to establish the
sequence of bond breaking and bond forming. The calculated anomeric carbon to former carbonyl oxygen bond that
breaks during acyl transfer is the longest in the formyl case and shortest in those that exhibit little or no acyl transfer.
Rotation about the aromatic to carbonyl Ph—C(=O) bond is different from the alkyl series. Analysis of this proposed
TS led to the postulate that 2,6-substitution may hinder rotation even more. Thus, the 2,6-dimethylbenzoyl analogue
was synthesized and it does not transfer directly or by rearrangement of its readily formed orthoester. DFT calculations
suggested that 2,6-dimethoxybenzoyl should also not transfer easily. Experimentally, this proved to be the case and this
new 2-O-acyl protecting group cleaves at 50 °C with a 1 mol/L solution of LiOH in methanol. Thus, a calculated tran-
sition state has led to a prototype of a protecting group that solves a major problem in oligosaccharide synthesis.
Key words: glycosylation, carbohydrates, quantum chemistry, reaction mechanism, neighboring-group effects. 1171
Résumé : Dans ce travail, on sépare sans équivoque et pour la première fois la formation d’orthoester des transferts
d’acyles et on suggère des voies possibles pour éliminer le transfert de 2-O-acyle lors des réactions de glycosylation.
On présente des données expérimentales montrant que le transfert d’acyle de donneurs glycosyles dérivés du 2-O-acyl-
3,4,6-tri-O-benzyl-D-galactopyranose diminue dans l’ordre formyle > acétyle > pivaloyle. Les dérivés 2-O-benzoyles
sont plus variables; dans certains cas, le transfert se fait facilement alors que dans d’autres il ne se fait pas du tout.
Des calculs basés sur la théorie de la densité fonctionnelle (THF) de la structure et des énergies de l’ion dioxolénium
et d’intermédiaires apparentés suggèrent qu’un transfert de proton du nucléophile vers O-2 s’avère la meilleure explica-
tion pour les tendances observées. Ces calculs de THF de la voie de transfert de proton appuient un mécanisme dans
lequel une molécule relais serait impliquée. Des calculs supplémentaires de THF ont été réalisés en appliquant une
contrainte basée sur des combinaisons linéaires de six longueurs de liaisons pour déterminer la séquence de bris et de
formation de la liaison. La longueur calculée pour la liaison entre le carbone anomérique et l’oxygène du carbonyle an-
térieur qui se brise au cours du transfert d’acyle est la plus longue dans le cas du formyle et elle est la plus courte
dans les cas de ceux qui ne donnent que peu ou pas de transfert d’acyle. La rotation autour de la liaison aromatique
vers le carbonyle de Ph—C(=O) est différente de celle observée dans la série alkyle. L’analyse de cet état de transition
proposé conduit au postulat que la substitution en 2 et 6 peut empêcher la rotation encore plus. On a donc réalisé la
synthèse de l’analogue 2,6-diméthoxybenzoyle et on a observé qu’il n’y a pas de transfert direct ou par le biais d’un
réarrangement de la part de cet orthoester facilement formé. Des calculs de TDF suggèrent que le 2,6-diméthoxybenzoyle
ne devrait pas donner lieu à des transferts. D’un point de vue expérimental, cette prédiction s’est avérée juste et ce
nouveau groupe protecteur 2-O-acyle peut être clivé à 50 °C, avec du 1 mol/L LiOH dans du méthanol. Ainsi, un état
Received 5 December 2003. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 17 September 2004.
A. Bérces.2 Steacie Institute for Molecular Sciences, National Research Council Canada, 100 Sussex Drive, Ottawa, ON K1A 0R6,
Canada.
D.M. Whitfield.3 Institute for Biological Sciences, National Research Council Canada, 100 Sussex Drive, Ottawa, ON K1A 0R6,
Canada.
T. Nukada. The Institute for Physical and Chemical Research (RIKEN), Wako-shi, 351-01 Saitama, Japan.
I. do Santos Z., A. Obuchowska, and J.J. Krepinsky. Department of Medical Genetics & Microbiology and Protein Engineering
Network of Centers of Excellence (PENCE), University of Toronto, Toronto, ON M5S 1A8, Canada.
1NRC paper No. 42487.
2Present address: Astra Zeneca R&D Mölndal, SE-431 83 Mölndal, Sweden.
3Corresponding author (e-mail: dennis.whitfield@nrc-cnrc.gc.ca).
Can. J. Chem. 82: 1157–1171 (2004)
doi: 10.1139/V04-059
© 2004 NRC Canada

1158
Can. J. Chem. Vol. 82, 2004
de transition calculé a conduit à un prototype de groupe protecteur qui résout un problème majeur dans la synthèse
d’oligosaccharides.
Mots clés : glycosylation, carbohydrates, chimie quantique, mécanisme réactionnel, effets de groupes voisins.
[Traduit par la Rédaction]
di-O-benzoyl-3,4-O-isopropylidene-1-thio-β-D-galactopyrano-
Introduction
Bérces et al.
side (2) under a wide variety of reaction conditions. Donor 2
was designed to allow for subsequent branching at O-3 and
O-4. Among the reaction products were: 3 the result of
benzoyl transfer to the nucleophile, 4 the expected product,
5 a 2-OH glycoside, and 6 β(1 → 2)-linked oligomers that
accounted for up to half the polymer bound products (see
Scheme 1 (10)). Previously (1 → 2)-linked disaccharides
have been isolated as side products in glycosylation reactions
that also exhibited acyl transfer (11). Also, some groups
have independently developed protocols to preparatively
prepare (1 → 2)-linked disaccharides based on this reaction
type (12). Since preparative polymer-supported chemistry
requires very high yields with no side products, we adopted
a program to develop strategies towards the elimination of
the acyl transfer side reaction.
Many biologically active molecules contain sugar mole-
cules with acetal or ketal linkages. The study of the di-
astereomeric complexities of joining two sugar molecules
together to form such linkages is an active area of research.
Most linkages are formed by glycosylation reactions be-
tween an activated electrophilic glycosyl donor and a weakly
nucleophilic alcoholic acceptor. This necessity to use highly
reactive donors makes these reactions prone to side reac-
tions. Minimizing side reactions and achieving stereochem-
ical control are the two main objectives of research into
glycosylation reaction mechanisms (1).
The commonest approach to stereochemical control of
glycosylation reactions is to use neighboring group partici-
pation, typically of an ester at C-2 of a hexopyranosyl donor.
After activation by a promoter in the absence of nucleo-
philes such species form dioxolenium ions by nucleophilic
attack of the ester carbonyl oxygen (O-7) on C-1 (2). It has
been shown theoretically that such cations have a LUMO
predominantly situated on the former carbonyl carbon (C-7)
and therefore C-7 should be the site of nucleophilic attack.
In the presence of nucleophiles, calculations suggest that sta-
ble intermediates that have long C-7—Onuc bonds are
formed (3). After shortening of this bond, hydroxylic proton
transfer and pyranose ring conformational change stable
orthoesters can be formed. A common side reaction long
associated with orthoester formation is acyl transfer. This
unwanted side reaction is the transfer of the O-2 acyl pro-
tecting group (4) to the nucleophilic acceptor to form a new
ester linkage and a free OH group at O-2 of the sugar (5).
This side reaction is the focus of this study.
It has long been known that increasing the steric bulk
about the acyl substituent minimizes acyl transfer. For exam-
ple, the suppression of acyl transfer during the preparation
of multigram quantitities of glucosylated sapogenins was
achieved by switching from peracetyl- to perbenzoyl-
substituted donors (13). Another group has shown the utility
of isobutyryl to minimize acyl transfer with glucuronyl do-
nors (14). Several groups have used pivaloyl esters to sup-
press acyl transfer (15). Most of these reports discuss the use
of pivaloyl in terms of suppressing orthoester formation and
do not discuss acyl transfer (16). Similarly, the mesitoyl
(2,4,6-trimethylbenzoyl) group has also been used to sup-
press orthoester formation (17). Orthoesters can be rear-
ranged to glycosides by acid catalysis (18) but acyl transfer
is frequently observed in this reaction (19). As we will show
below, the propensity to form orthoesters and the propensity
to give acyl transfer are not directly related (20).
Acyl transfer from the glycosyl donor to the sugar accep-
tor has been noted numerous times over many years (6).
This side reaction has been observed with many different
glycosyl donors (7) under many different glycosylation reac-
tion conditions. Besides the acylated acceptor alcohols,
glycosides often of both anomers with a 2-OH group, for ex-
ample 5, are frequently formed as well (8). For “conven-
tional” solution-phase chemistry this side reaction is only a
problem of yield and adds extra complexity to the reaction
mixture to be purified. For solidphase methodologies, where
the acceptor is bound to the polymeric support, it is particu-
larly troublesome (9). If the acyl group is used as the
cleavable group then two or more hydroxyl species instead
of one will be available for the subsequent glycosylation. If
the acyl group is not the cleavable protecting group then the
acyl transfer itself essentially caps the polymer bound accep-
tor and it is the formation of polymer bound donor derived
glycosides with free OH groups that can undergo further
glycosylation, which is the problem.
Results and discussion
Few studies have attempted to systematically study the ef-
fect of the acyl group on acyl transfer. We had previously
studied a series of 2,3,4,6-tetra-O-acyl-substituted galacto-
pyranosyl trichloroacetimidate donors (7ab and 8), and
shown that the order of acyl transfer was benzoyl (12b) <<
isobutyryl (12c) ≈ acetyl (12a) (8). This study also showed
that trifluoromethanesulfonic anhydride as promoter with the
peracetate 7a (21) led to 1,2 linked-oligomers, whereas us-
ing silver or copper(II) triflate as promoter eliminated the
1,2-oligomers but not acetate transfer (see Scheme 2). We
now show that the powerful promoter triethylsilyltriflate also
leads to acetate transfer in a ratio of 40:60 (glycoside 11a to
acetate transfer 12a) for the α-trichloroacetimidate 7a and
33:67 for the β-trichloroacetimidate 9a. Changing to the
perbenzoyl trichloroacetimidate 9b improves the ratio to
93:7 (see Experimental). This matches the previously re-
ported result with the perbenzoyl 1-thio-galactoside donor
Such a reaction was observed between the linker-polymer
combination MPEG-DOXOH (1) and the donor ethyl 2,6-
© 2004 NRC Canada

Bérces et al.
1159
Scheme 2. Effect of acyl groups on acyl transfer with galactose
Scheme 1. Acyl transfer side reaction during a polymer-supported
glycosylation reaction. Even after extensive optimization desired
glycoside 4 could only be formed in <40% yield (10).
donors (8, 10).
ion B is derived by ionization of donor A. Monocyclic B af-
ter ring inversion can cyclize to the bicyclic dioxolenium ion
C associated with neighboring group participation with a
calculated barrier of 34 kJ mol–1 (23). This ion has a vacant
p-like orbital centered on the former carbonyl carbon (C-7)
(3). This reacts with methanol to form an ion–dipole com-
plex D with a C-7—O-8 (O-8 from MeOH) bond length of
2.81 Å. After considerable search, a pathway involving pro-
ton transfer to O-2 from D spontaneously led to acyl transfer
and the 2-OH ion E.4 Ion E is presumably the precursor to
2-OH glycosides and 1,2-oligomer formation but its reactiv-
ity has not been studied further.
To further study the acyl transfer pathway, we decided to
study the effect of acyl substituents. For this purpose, a
series of 2-O-acyl 3,4,6-tri-O-benzyl-D-galactopyranosyl do-
nors (13a–13e), where the acyl groups are formyl, acetyl,
benzoyl, pivaloyl, and levulinoyl, were prepared and reacted
with 1 (see Supplementary material5 and Experimental. The
ease of acyl transfer during the preparation of glycosides
14a–14e followed the order formyl > acetyl > levulinoyl >>
benzoyl = pivaloyl in the amounts 56%, 5%, 2%, and 0%
(cf. 15a-15e, see Scheme 4). This compares to the 3,4-
isopropylidene donors where benzoyl transfers more than
pivaloyl with the same acceptor suggesting that the more re-
active 3,4,6-tri-O-benzyl donors suppress acyl transfer (24).
To study the acyl group effect theoretically we calculated
the intermediates B, C, D, and E′ for the series 2-O-acyl 3,4-
O-isopropylidene-6-O-acetyl-D-galactopyranosyl cation plus
MeOH where the acyl is formyl, acetyl, benzoyl, and
pivaloyl (see Table 1). Note as shown below, E′ in this series
is an ion–dipole complex of the ion E and the appropriate
acyl methyl ester. The pyranose ring in all C and Ds is in the
10b, where a ratio of 93:7 (11b:12b) was also found (10).
We interpret these results to indicate that one or more inter-
mediates are formed in such glycosylation reactions that
eventually lead to acyl transfer or glycosylation (cf. D and E
in Scheme 3).
Returning to 3,4-O-isopropylidene galactose derivatives,
we showed that even pivaloyl esters transfer to form MPEG-
DOXOPiv although 1,2-oligomer formation was suppressed.
Following a literature precedent (22), increasing the steric
bulk about the alcohol acceptor by forming the α-methyl
benzyl derivative MPEG-MDOXOH was undertaken for the
preparation of synthetically useful amounts of MPEG-
MDOXyl glycosides. Two other factors, namely operating at
the highest possible concentration of TfOH and the highest
possible temperature that did not lead to decomposition re-
actions, were also necessary to suppress acyl transfer. These
last two observations can be interpreted to indicate that acyl
transfer gives the apparent kinetic product and glycosylation
the apparent thermodynamic product (10). The importance
of the acid concentration also suggests that proton transfer is
integral to the mechanism as previously proposed (11a).
Given the high propensity to undergo acyl transfer of the
3,4-isopropylidene galactose derivatives, we undertook to
study the mechanism of acyl transfer with the 2,6-di-O-
acetyl-3,4-O-isopropylidene-D-galactopyranosyl cations. The
structures and energetics of the following intermediates were
calculated using density functional theory (DFT) with a con-
tinuum solvent correction (see Scheme 3). The “parent” cat-
4 In the original report of ref. 3 other ion–dipole complexes resulting from the methanol approach to the anomeric center of C, to the β-face
of B, and the α-face of B, respectively, as well as a plausible intermediate with a tetrahedral carbon were considered.
5 Supplementary data may be purchased from the Depository of Unpublished Data, Document Delivery, CISTI, National Research Council
Canada, Ottawa, ON K1A 0S2, Canada (http://www.nrc-cnrc.gc.ca/cisti/irm/unpub_e.shtml for information on ordering electronically).
© 2004 NRC Canada

1160
Can. J. Chem. Vol. 82, 2004
Scheme 3. Previously proposed mechanism for acyl transfer via proton transfer to O-2. Intermediates B–E derived by the ionization of
donor A (10).
and pivaloyl (1.618 Å). These trends fit the observation of
acyl transfer diminishing in the series formyl > acetyl >
pivaloyl (see Figs. 1a–1c). The benzoyl D has the shortest
calculated C-1—O7 bond length (1.607 Å) but is quite close
to pivaloyl (cf. Figs. 1d and 1c). These results suggest some
π-resonance contribution, which is not unexpected given the
similarity to a substituted benzyl cation. Similarly, the
benzoyl complex D is the most stable relative to isolated B
and MeOH as shown in Table 1.
Scheme 4. 2-O-Acyl-substituted galactose donors used to assess
the substituent effect on acyl transfer.
From our previous calculations we know that proton
transfer triggers the acyl transfer and for this reason, the pro-
ton transfer is probably the rate-determining step (10). We
first studied the direct transfer of the H-(O-8) to O-2 using
constrained linear transit calculations explained in Computa-
tional details. We found that the direct proton transfer in-
volves a 170 kJ mol–1 barrier, which is clearly too high and
this mechanism is therefore unrealistic. For a proton transfer
to take place with a reasonably low energy barrier, the do-
nor, the acceptor, and the proton should ideally be situated
on a straight line. In contrast, the angle between the H—O-8
bond and the H—O-2 hydrogen bond is 78° in the formyl
case. Consequently, the proton transfer does not take place
directly but likely involves a relay molecule. We note that
these reactions normally take place in aprotic solvent but a
small amount of water or other proton relay molecules can-
not be ruled out. The counteranions or leaving-group derived
entities as well as added bases and molecular sieves etc. are
all possible relay molecules. To keep the model as simple as
possible we considered a second methanol to be involved as
2
B2,5 conformation and in all Bs is in an SO conformation.
Figure 1 shows partial ball and stick representations of the
structures of the D complexes. It is readily apparent that the
C-1—O-7 bond length is the longest for the formyl deriva-
tive among the series formyl (1.810 Å), acetyl (1.640 Å),
© 2004 NRC Canada

Bérces et al.
1161
Table 1. Relative energies in kJ mol–1 for cations B–E′ derived from 2,6-di-O-acyl-3,4-O-isopropylidene-D-galactopyranosyl donors.
Formyl
Acetyl
Pivaloyl
Benzoyl
2,6-Dimethylbenzoyl
2,6-Dimethoxybenzoyl
0.0a
–34.9
–35.9
–44.6
0.0b
0.0
–51.6
–57.7
–46.2
0.0
0.0
–48.7
–56.1
–16.6
0.0
0.0
–56.0
–62.6
–30.8
0.0
0.0
–65.9
–70.0
–46.9
0.0
0.0
–82.6
–85.2
–57.6
0.0
B
C
D
E′
ME (c)
ME (t)
–10.7
–18.3
–44.8
–33.5
–15.7
–12.8
aEnergy including solvated methanol for B and C at the same level of theory.
bEnergy for the solvated cis (c) ester set to 0.0 for comparison to the trans ester (t) reaction (25).
Fig. 1. Partial ball and stick representations of ion–dipole complexes D for: (a) formyl, (b) acetyl, (c) pivaloyl, (d) benzoyl, (e) 2,6-
dimethylbenzoyl, and (f) 2,6-dimethoxybenzoyl.
© 2004 NRC Canada

1162
Can. J. Chem. Vol. 82, 2004
Scheme 5. The constraint based on r1r6 used for the mechanism
calculation (see eq. [2]).
events, whether the bond breaking or the formation takes
place first or if the two steps are synchronized. In addition,
it is straightforward to extend this constraint to the case de-
scribed in Scheme 5. For the exploration of the reaction path
from F to G we used the following definition of the con-
straint:
[2]
Q = r1 – r2 + r3 – r4 + r5 – r6
where r1 through r6 are defined in Scheme 5. In our calcula-
tions, we varied Q stepwise from a negative value to a posi-
tive one in 50–60 steps. At each value of the constraint the
geometry was optimized subject to the constraint. Conse-
quently, going from one value to the next of Q in the reac-
tion path, the bond that changes its length is always
whichever bond involves the least increase in the energy.
These calculations enable one to find out the sequence of
events without any assumption. However, this constraint has
its own limitations. If the mechanism is stepwise, after the
first proton transfer the broken bond involving the trans-
ferred hydrogen becomes the softest coordinate. Conse-
quently, the lowest energy path involving the increase of Q
is the increase of the distance of the broken bond, which
may be irrelevant to the mechanism. For this reason, in a
clearly stepwise reaction, once the first step is explored Q
may be modified to the exploration of the next step to ex-
clude the coordinates irrelevant for the rest of the reaction.
In addition, in the definition of Q we assumed that all
conformational changes are the consequence of the forma-
tion and breaking of the bonds. As we found in the calcula-
tions, the products often involve different conformations
from the reactants. To explore the importance of conforma-
tional change in the mechanism in a quantitative manner is a
subject for future research. Even with its limitations, these
calculations gave some significant insight into the reaction
mechanism. It is remarkable that out of the 720 theoretically
possible reaction mechanisms, all studied reactions follow
essentially the same steps with four different protecting
groups (see Fig. 2a–2f for the acetyl example). The potential
energy surface involves an initial barrier followed by a high
energy intermediate in a plateau region (see Fig. 3). The pla-
teau is followed by a second barrier, which is higher than the
first barrier except for the pivaloyl case.
The mechanism starts with the shortening of the O-8—C-7
bond length, which markedly increases the acidity of the
nucleophilic hydroxlic proton. This charge movement pre-
sumably triggers transfer of this hydroxylic proton from O-8
to the relay molecule (see Fig. 2b) followed immediately by
the transfer of the proton from the relay molecule to O-2.
The first barrier (TS1) on all calculated potential surfaces
corresponds to the second proton transfer step to O-2 (see
Fig. 4). The first proton transfer causes the formation of a
shoulder on the potential surface but no barrier can be asso-
ciated with it. The intermediate is not clearly defined due to
the shallow nature of the plateau region. The second barrier
(TS2) involves the breaking of the O-2—C-7 bond and
charge transfer to C-7 associated with the change of sp3 to
sp2 hybridization of C-7. Subsequently the C-1—O-7 bond
breaks to give G. This mechanism resembles the intuitive
proposal of Bochkov et al. (26) except that their model has
a relay molecule in the proton transfer. Similar relay mole-
cules have been used in theoretical studies of the mutaro-
tation reaction (25).
To understand the mechanism of acyl transfer we carried
out calculations mapping the reaction path leading from F to
G as shown in Scheme 5. General methods of reaction path
following usually consider elementary reaction steps. The
application of these methods becomes intractable for multi-
step reactions such as acyl transfer. Consequently, we cus-
tomized our approach to this reaction to gain insight into the
mechanism. We carried out linear transit calculations using a
constraint that ensures that the reactant and the product
states correspond to F and G.
This reaction involves the formation and breaking of
seven bonds as well as hybridization changes such as C-1
going from sp2 in F to sp3 in neutral products. Two hydro-
gen bonds become covalent bonds and two covalent bonds
involving hydrogen break to become hydrogen bonds. In ad-
dition, the formation of the O-8—C-7 bond and the breaking
of O-2—C-7 and O-7—C-1 are key features in the mecha-
nism. Since the O-7—C-1 bond breaking is the consequence
of the rest of the rearrangements, we do not have to consider
this bond as part of our linear transit constraints. To consider
the breaking and formation of six separate bonds individu-
ally requires assessing the 6 × 5 × 4 × 3 × 2 = 720 possible
sequences of events. For this reason, we have to define a
constraint that brings the reactants to product state without
predefining the sequence of events. We shall explain how
this can be carried out on a simpler example and extend the
results to acyl transfer. Lets consider the transfer of H—(O-
R) to O-2. There are two bonds involved in this transfer: the
forming H—O-2 bond and the breaking H—(O-R) bond. A
simple constraint
[1]
Q = R[H—(O-R)] – R[H—(O-2)]
where R[H—(O-R)] and R[H—(O-2)] are the H—(O-R) and
H—(O-2) bond lengths, respectively, can be used to study
this mechanism. At the reactant state, the value of Q is nega-
tive while it is positive in the product state. To study this re-
action by quantum mechanical methods, we can carry out
constrained geometry optimization at varying values of Q
representing the progress of the reaction. This procedure is
called linear transit. The application of Q as a constraint has
some distinct advantages compared to using either the form-
ing or the breaking bond distance as constraint alone. Most
importantly, this constraint does not specify the sequence of
© 2004 NRC Canada

Bérces et al.
1163
Fig. 2. Ball and stick representations of the constrained acyl transfer reaction for the acetyl case. (a) Initial structure showing relay
methanol (OR and OHR). (b) Proton transfer to the relay methanol. (c) Proton transferring from the relay methanol to O-2 (1st TS).
(d) Proton transferred to O-2. (e) Breaking of bond O-2—C-7 (2nd TS). (f) Final structure with bond C-1—O-7 broken also.
an equivalent to E′ with a covalent O-7—C-1 bond and no
proton relay.
tion of the final ester product. Methyl pivalate is found in
the preferred trans conformation, while all other esters were
found in the undesirable cis conformation at the end of the
reaction path (see Figs. 2f and 5). The steric repulsion of the
bulky pivaloyl group prevented the rotation of the methyl
group towards the cis orientation. For all other systems, the
rotation of the methyl group is easier towards the cis
position at the beginning of the reaction. Consequently, we
consider the large second barrier an artifact of the confor-
mational orientation. This clearly points to the need to im-
prove the reaction path following method by enabling the
The first barrier is between 30 and 78 kJ mol–1, the formyl
being the lowest and the benzoyl the highest. The second
barrier is between 40 and 92 kJ mol–1, the pivaloyl being the
lowest and the benzoyl the highest. The second barrier ap-
pears as a sharp peak in all potential surfaces except the
pivaloyl (see Fig. 3). We found that the lack of the large
peak at the second barrier in the case of the pivaloyl is due
to a qualitatively different pathway. The difference is best
understood by considering the difference in the conforma-
© 2004 NRC Canada

1164
Can. J. Chem. Vol. 82, 2004
Fig. 3. Bond lengths (r1–r6) used for the constrained acyl transfer reaction plus C-1—O-7 and the relative energy vs. the constraint
step. The acyl group is pivaloyl.
involvement of conformational as well as bond distance co-
ordinates in the constraints. Such a development is for future
studies.
The above analysis suggested to us that in the benzoyl
case 2,6-disubstitution may inhibit this rotation and hence
acyl transfer. The DFT-calculated intermediates B–E′ are
shown in Table 1 and Fig. 1e. Like the parent benzoyl, the
2,6-dimethylbenzoyl D is predicted to be stable with a short
C-1—O-7 bond length. Thus, we synthesized the 2,6-
dimethylbenzoic acid analogue in the 3,4-isopropylidene se-
ries and tested its reactivity (see Experimental and
Scheme 6). This derivative 17 synthesized from known diol
16 (27) exhibited no acyl transfer giving an almost quantita-
tive yield of glycoside 18. The structure of 18 was
confirmed by cleaving from the polymer to give 19 (28).
Furthermore, if less triflic acid was added (0.7 equiv. vs.
1.4 equiv.) then a mixture of orthoester 20 and 18 was ob-
tained. The orthoester 20 in the mixture could be smoothly
isomerized to 18 with more triflic acid without acyl transfer.
To the best of our knowledge this is the first experimental
result that uncouples orthoester formation and acyl transfer.
Orthoesters (29) have been previously suggested to be in-
termediates to glycosides and acyl transfers (30). The acid-
catalyzed isomerization of 20 to 18 without acyl transfer
strongly supports a mechanism for acyl transfer that does not
involve neutral orthoesters. Note that our D ion–dipole com-
plexes have long C-7—O-8 bond lengths (see Figs. 1a–1f),
and must undergo three major changes to become neutral
orthoesters: the O-8—C-7 bond must shorten, the pyranose
ring conformation must change, and the alcoholic proton
must be transferred. DFT calculations (not shown) of this
process suggest that proton transfer is the last of these three
steps. This uncoupling of orthoester formation and acyl
Taking all these finding into consideration, the rate-
determining barrier is the first barrier on the calculated reac-
tion surface associated with proton transfer to O-2. The
height of this barrier is 30, 40, 40, and 78 kJ mol–1 for
formyl, acetyl, pivaloyl, and benzoyl, respectively. Some
contribution from the undesirable conformation can be at-
tributed to the first barrier in all cases except the pivaloyl.
The largest difference between the cis and trans product was
found in the case of benzoyl ester, which is 33.5 kJ mol–1,
while it is 10.7 and 18.3 in the case of formyl and acetyl
(see Table 1). Considering the relative magnitudes of these
undesirable contributions, our calculated barriers are consis-
tent with the experimentally found order of propensity to
acyl transfer.
From these four trajectories we can draw some important
conclusions. The conformational reorganization is driven by
the change in the hybridization of C-7, which goes from sp3
to sp2, and consequently from tetrahedral to a planar confor-
mation. In addition to the conformational change leading to
either cis or trans in the final ester product, the protecting
groups also play a role in determining the conformation
around the C-7—O-7 bond, which is crucial in the reaction
mechanism (see Fig. 5). In this regard, the benzoyl stands
out as different from the rest and especially different from
pivaloyl. The flat benzoyl group can accommodate a confor-
mation helping the C-7—O-2 bond breaking by simply rotat-
ing around the C-7—phenyl bond.
© 2004 NRC Canada

Bérces et al.
1165
Fig. 4. Ball and stick representations of putative TS1: (a) formyl, (b) acetyl, (c) pivaloyl, (d) benzoyl.
transfer is in agreement with the ease of formation of
pivaloyl orthoester (31) but their reluctance to undergo acyl
transfer. What the mechanism is for proceeding from D to
glycosides is not known (32). This is a subject of current re-
search for us (33).
92%:8%). We had also envisaged that the 2,6-dimethoxy an-
alogue would cleave easier than the 2,6-dimethyl one due to
the possibility of chelation to the methoxy oxygens. The 2,6-
dimethoxybenzoyl group is smoothly cleaved with 1 mol/L
LiOH in methanol at 50 °C for 16 h to diol 24. To improve
the ease of removal, we are examining 2,6-dimethoxy-4-X-
benzoyl esters, where X is an electron-withdrawing group.
Thus, we have progressed from a calculated TS to a proto-
type of a protecting group that minimizes acyl transfer.
Since the 3,4-isopropylidene galactose donors are the most
prone to acyl transfer among those that we have studied, this
new protecting group should be a general solution to acyl
transfer.
Our previous results had shown that even benzoate is dif-
ficult to cleave from polymer-supported Gal C-2 (10) and it
is well-known that 2,6-dimethylbenzoyl esters are difficult to
hydrolyse (34). Therefore, it was not surprising that under a
variety of vigorous conditions (for example LiAlH4 at reflux
in THF, 1 mol/L KOHaq at 50 °C, etc.), we could not cleave
the 2,6-dimethylbenzoate group from 18 without decomposi-
tion or recovery of intact 18. Eventually we discovered that
1 mol/L LiOH in methanol at 70 °C for 48–60 h did cleave
this group (see Scheme 7). This led us back to the results in
Figs. 1e and 1d where the C-1—O-7 bond length (1.584 Å)
in complex D of the 2,6-dimethylbenzoate is shorter than in
D of the benzoate (1.607 Å). This suggested to us that the
difference in this bond length may be at least in part a π-
electronic effect. Therefore, we calculated the parameters for
the more electron-donating 2,6-dimethoxybenzoate analogue,
and the C-1—O-7 bond length (1.557 Å) is even shorter (see
Fig. 1f). Subsequently, we synthesized this analogue 21 from
16 and reacted it with 1 (see Scheme 7). The reaction pro-
ceeded smoothly to give glycoside 22 contaminated with
only a small amount of acyl transfer product 23 (22:23,
Conclusions
Computational evidence is presented that ion–dipole com-
plexes like D are formed under glycosylation reaction condi-
tions. A pathway from D that most likely involves a proton
relay from the nucleophile’s proton to O-2 of the electro-
philic sugar with accompanying formation of the O-8—O-7
bond, O-2—C-7 bond rupture, and finally O-7—C-1 bond
breaking leads to an ion–dipole complex of a 2-OH glycosyl
cation and the ester by-product E′. This order of bond form-
ing and bond breaking was determined by computations us-
ing a complex constraint (see Scheme 5). In the model, the
© 2004 NRC Canada

1166
Can. J. Chem. Vol. 82, 2004
Fig. 5. Ball and stick representations of putative TS2: (a) formyl, (b) acetyl, (c) pivaloyl, (d) benzoyl.
ester is formed in the unfavorable cis conformation except
the pivaloyl, which perhaps explains the strong sensitivity to
steric bulk of the acyl transfer side reaction. The O-7—C-1
bond in the D complexes is the longest in the formyl case
and the shortest in the pivaloyl case along the series formyl,
acetyl, pivaloyl. It is even shorter in the benzoyl case even
though benzoyl still transfers in some cases suggesting an
electronic component besides a steric component to this side
reaction. The steric effect led us to use the 2,6-dimethyl- and
2,6-dimethoxybenzoyl groups as neighboring groups, which
eliminate or greatly minimize acyl transfer especially com-
pared to benzoate where >50% of the reaction products were
1,2-linked oligomers under the same reaction conditions.
The 2,6-dimethoxybenzoyl is easier to remove and its use is
under investigation in other oligosaccharide syntheses.
Orthoester formation with the 2,6-dimethylbenzoyl protect-
ing group is facile but its rearrangement to glycoside is not
accompanied by acyl transfer. Although these two side reac-
tions are probably related, this result clearly shows that the
two reactions are distinct. This separation of reaction mecha-
nisms strongly suggests that acyl transfer can be eliminated
by the judicious choice of protecting groups and reaction
conditions like the ones described here.
done under argon. The progress of the reactions was moni-
tored by thin layer chromatography (TLC) on silica gel. The
TLC results were visualized under UV light (254 nm) and
by spraying with 50% sulphuric acid in methanol and heat-
1
ing at 200 °C. The H and 13C NMR spectra were recorded
in deuteriochloroform solution at 500.1 MHz and
125.8 MHz, respectively, with Varian Unity Plus spectrome-
ters. For polymer-bound samples, the MPEG methylenes were
saturated and quantitation was made by comparing integrals
to the terminal methyl of the MPEG. Assignments were
made by comparison to the spectra of building blocks and
1
cleaved compounds. H NMR spectra in CDCl3 were refer-
enced to residual CHCl3 at 7.26 ppm, and 13C NMR spectra
to the central peak of CDCl3 at 77.0 ppm. Assignments were
1
made by standard H–1H-COSY and 13C-1H-COSY experi-
ments. The mass spectral analysis was done on a forward
mass spectrometer. Fast atom bombardment (FAB) MS was
performed using xenon atom at 6 kV as the source. Thio-
glycerol or a mixture of glycerol and thioglycerol was used
as FAB matrix. Typically, 10–15 full-range, low-resolution
MS scans were averaged to yield a low-resolution mass
spectrum. For high-resolution MS, the electric sector was
scanned over the range of interest. Typically, polyethylene
glycol or polypropylene glycol was used as an internal mass
standard and between 75 and 150 scans were averaged.
Some MS were done by ion spray with a PerkinElmer/Sciex
API triple quadrupole mass spectrometer.
Methods
All starting materials were dried overnight in vacuo at
10–3 mmHg (1 mmHg = 133.322 4 Pa). All reactions were
1
The H NMR spectrum of the acylated MPEGDOXOH
© 2004 NRC Canada

Bérces et al.
1167
Scheme 6. 2,6-Dimethylbenzoyl as a protecting group to elimi-
nate acyl transfer and to separate acyl transfer from orthoester
formation.
were obtained by acylating the polymer i.e., either formyl,
acetyl, or benzoyl. The H NMR spectrum of MPEGDOX-
1
O-formyl shows that the CHO-O-CH2-C6H4-CH2-
(OCH2CH2)nOCH3 peak is shifted to 5.19 ppm from
4.68 ppm in MPEGDOXOH. This same peak appears at
5.09, 5.05, 5.10, 5.35, and 5.34 ppm in the acetylated (12a),
pivaloylated (15d), levulinoylated (15e), benzoylated (3),
and 2,6-dimethoxybenzoylated (23), MPEGDOXOH, re-
spectively.
Computational details
The Amsterdam density functional (ADF) calculations use
the methods described in ref. 3 and include a continuum di-
electric solvent term. The basis set used was a double ζ basis
set with a single polarization function. The conformational
description based on the IUPAC nomenclature is described
in detail in ref. 35. A fully working version of the program
is available at http://www.sao.nrc.ca/ibs/6ring.html.
To enable the exploration of the reaction pathways we in-
corporated some changes in the ADF program. ADF enables
the user to follow the reaction pathway by slowly varying a
constrained coordinate from the value representative of the
reactant state to the product state with geometry optimiza-
tion subject to this constraint at given values of the con-
straint. This procedure is called a linear transit calculation.
The original implementation of linear transit allows only in-
dividual internal coordinates such as bond length, bending
angles, and torsion angles to be defined as a constraint.
However, it is not possible to define linear combinations of
individual internal coordinates, which is desirable for com-
plicated reaction profiles such as acyl transfer. To facilitate
the exploration we extended the definition of constraint by
implementing a general constraint of the following form:
Scheme 7. 2,6-Dimethoxybenzoyl as a protecting group for mini-
mizing acyl transfer.
Q = l1*q1 + l2*q2 … + lnqn
where q can be any bond length, bending angle, or torsion
angle.
The coordinates are kept constant in the geometry optimi-
zation by simply projecting out the contribution of the Carte-
sian forces to the force on the constrained coordinate
followed by a Cartesian geometry optimization step. After
the Cartesian optimization step, the constraints are exactly
imposed by an iterative correction of the coordinates (36).
MPEGDOXOH (1) was synthesized according to ref. 37.
The synthesis of donors 13a–13e is described in the Supple-
mentary material.5
MPEGDOXyl 2,3,4,6-tetra-O-acetyl-␤-D-
galactopyranoside 11a (38)
Experiment I (using ␣-imidate)
A solution of TESOTf in CH2Cl2 (2.0 µL, 8.8 µmol) was
added to a solution of 2,3,4,6-tetra-O-acetyl-α-D-galacto-
pyranosyl trichloroacetimidate (39) (8a) (27 mg, 55 µmol)
and 1 (57 mg, 27 µmol) in dry CH2Cl2 (2 mL) at room tem-
perature. After 1 h of stirring, TLC (hexanes : ethyl acetate,
2:1, 2% diisopropylethylamine (DIPEA)) indicated complete
disappearance of the starting imidate. The reaction was
stopped 3 h after the addition of TESOTf by addition of
© 2004 NRC Canada

1168
Can. J. Chem. Vol. 82, 2004
NaHCO3 (and stirring for ~10 min), filtered, and the volume
was reduced to ~0.5 mL. Dry tert-butylmethylether (TBME)
(30 mL) was then added and the polymer was precipitated at
4 °C overnight. The polymer (64 mg) was then isolated by
filtration. The 1H NMR spectrum indicates that 40% yield of
the β-glycosylated product 11a and 63% yield of acyl trans-
J1,2 = 7.9 Hz, J2,3 = 10.0 Hz, H-2), 4.37 (d, 1H, H-1), 3.94
(m, 1H, H-4), 1.99 (s, 3H, CH3CO).
MPEGDOXyl 2-O-benzoyl-3,4,6-tri-O-benzyl-
␤-D-galactopyranoside (14c)
1
fer product 12a were obtained. H NMR (ppm) δ: 5.39 (dd,
Compound 13c (47 mg, 67 µmol) was treated as for 14a.
1
1H, J3,4 = 3.4 Hz, J4,5 = 1.0 Hz, H-4), 5.27 (dd, 1H, J1,2
=
The H NMR spectrum indicated that 66% yield of the β-
8.1 Hz, J2,3 = 10.5 Hz, H-2), 4.98 (dd, 1H, H-3), 4.51 (d,
1H, H-1), 4.21 (dd, 1H, J5,6 = 6.4 Hz, J6,6′ = 11.3 Hz, H-6),
4.16 (dd, 1H, J5,6′ = 6.8 Hz, H-6′), 3.89 (m, 1H, H-5), 2.16
(s, 3H, CH3CO), 2.07 (s, 3H, CH3COO), 2.02 (s, 3H,
CH3CO), 1.98 (s, 3H, CH3CO).
glycosylated product 14c, 11% yield of the orthoester, and
1
no detectable acyl transfer product were obtained. 14c: H
NMR (ppm) δ: 7.98–7.96 (m, 2H, Bzo), 7.59 (m, 1H, Bzo),
7.47–7.43 (m, 2H, Bzp), 5.71 (dd, 1H, J2,1 = 8.0 Hz, J2,3
=
9.8 Hz, H-2), 4.00 (m, 1H, H-4). Orthoester: 5.25 (d, 1H, H-
1).
Experiment II (using ␤-imidate)
TESOTf (3.9 µL, 17 µmol) was added to a solution of
2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl trichloroacetimi-
date 9a (43 mg, 87 µmol) and 1 (92 mg, 43 µmol) and were
reacted as above to yield 33% of the β-glycosylated product
11a and 67% of acyl transfer product 12a.
MPEGDOXyl 3,4,6-tri-O-benzyl-2-O-pivaloyl-
␤-D-galactopyranoside (14d)
Compound 13d (17 mg, 25 µmol) was treated as for 14a.
1
The H NMR spectrum indicated that 47% yield of the β-
glycosylated product 14d, 14% yield of the orthoester, and
1
no detectable acyl transfer product were obtained. 14d: H
MPEGDOXyl 2,3,4,6-tetra-O-benzoyl-␤-D-
galactopyranoside (11b)
NMR (ppm) δ: 5.49 (dd, 1H, J2,1 = 7.9 Hz, J2,3 = 10.0 Hz,
H-2), 4.42 (d, 1H, H-1), 3.94 (m, 1H, H-4).
2,3,4,6-Tetra-O-benzoyl-β-D-galactopyranosyl trichloroace-
timidate (39) (10b) (40 mg, 54 µmol) and 1 (57 mg,
27 µmol) were reacted as for 11a to yield 93% of the β-
glycosylated product 11b and 7% of acyl transfer product 3.
1H NMR (ppm) δ: 8.10 (d, 2H, J = 7.5 Hz, Bzo), 8.05 (d, 2H,
J = 7.5 Hz, Bzo), 7.92 (d, 2H, J = 7.7 Hz, Bzp), 7.78 (d, 2H,
J = 7.7 Hz, Bzp), 5.98 (d, 1H, J3,4 = 3.0 Hz, H-4), 5.86 (dd,
1H, J1,2 = 8.1 Hz, J2,3 = 10.1 Hz, H-2), 5.54 (dd, 1H, H-3),
4.29 (m, 1H, H-5).
MPEGDOXyl 3,4,6-tri-O-benzyl-2-O-
levulinoyl-␤-D-galactopyranoside (14e)
Compound 13e (23 mg, 33 µmol) was treated as for 14a.
1
The H NMR spectrum indicated that 49% yield of the β-
glycosylated product 14e, 7% yield of the orthoester, and
2% yield of acyl transfer product 15e were obtained. 14e: 1H
NMR (ppm) δ: 5.42 (dd, 1H, J2,1 = 7.9 Hz, J2,3 = 10.0 Hz,
H-2), 3.93 (m, 1H, H-4), 2.71–2.66 (m, 2H, CH2 Lev), 2.53–
2.49 (m, 2H, CH2 Lev).
MPEGDOXyl 3,4,6-tri-O-benzyl-2-O-formyl-␤-
D-galactopyranoside (14a)
Ethyl 2,6-di-O-(2,6-dimethylbenzoyl)-3,4-O-
isopropylidene-1-thio-␤-D-galactopyranoside
(17)
TESOTf (12 µL, 53 µmol) was added to a solution of
compound 13a (0.169 g, 0.271 µmol) and 1 (0.275 g,
0.130 µmol) in dry CH2Cl2 (3 mL). After 3 h of stirring at
room temperature, TLC (hexanes : ethyl acetate, 3:1) indi-
cated absence of the imidate. The reaction was stopped by
the addition of 2 drops of DIPEA and the volume was re-
duced to ~0.5 mL. Dry TBME (50 mL) was added and the
polymer was allowed to precipitate at 4 °C overnight. The
2,6-Diol (27, 40) (16) (1.5 g, 5.71 mmol) was dissolved in
THF (30 mL) and 2,6-dimethylbenzoic acid (3.43 g, 4
equiv.), dicyclohexylcarbodiimide (5.3 g, 4.5 equiv.), and
DMAP (50 mg) were added. After stirring at room tempera-
ture (RT) for 1 day, the mixture was heated to 40 °C and the
temperature maintained for 1 week. Each day, further
aliquots of DCC (0.5 equiv.) and DMAP (50 mg) were
added until the starting material disappeared. After filtration
and concentration, the residue was purified by MPLC eluting
with hexanes : ethyl acetate : dichloromethane (8:1:1) to
yield slightly impure material, which was recrystallized from
ethanol:water about 1:1 to form colorless needles of 17
(750 mg, 25%); mp 159 to 160 °C. [α]D = 35.5 (c 0.84,
1
polymer (0.278 g) was then isolated by filtration. The H
NMR spectrum indicated that ~25% yield of the β-
glycosylated product 14a, ~4% yield of the orthoester, and
56% yield of acyl transfer product 15a were obtained. 14a:
1H NMR (ppm) δ: 8.14 (s, 1H, HCOO), 5.10 (dd, 1H, J1,2
=
8.3 Hz, J2,3 = 9.0 Hz, H-2), 3.96 (d, 1H, J3,4 = 2.4 Hz, H-4).
1
CHCl3). H NMR (ppm) δ: 7.22 (m, 2H, Bz), 7.06 (m, 4H,
MPEGDOXyl 2-O-acetyl-3,4,6-tri-O-benzyl-␤-
Bz), 5.37 (brt, 1H, H-2), 4.65 (m, 2H, H-6ab), 4.45 (d, 1H,
J1,2 = 10.3 Hz, H-1), 4.31 (m, 2H, H-3, H-4), 4.17 (t, 1H,
J4,5+5,6 = 12.2 Hz, H-5), 2.78 (m, 2H, CH3CH2S), 2.43, 2.37
(2s, 12H, ArCH3), 1.66, 1.39 (2s, 6H, O2C(CH3)2), 1.28 (t,
3H, J = 7.3 Hz, CH3CH2S). 13C NMR (ppm) δ: 169.6, 168.6
(2COAr), 135.3, 135.1 (2Bzip), 133.4 (2Bzo), 129.5, 129.4
(2Bzp), 127.5, 127.5 (2Bzm), 110.8 (O2C(CH3)2), 82.2 (C-1),
D-galactopyranoside (14b)
Compound 13b (27 mg, 42 µmol) was treated as for 14a.
1
The H NMR spectrum indicated that 64% yield of the β-
glycosylated product 14b, 13% yield of the orthoester, and
5% yield of acyl transfer product 15b were obtained. 14b:
1H NMR (ppm) δ: 7.36–7.22 (m, aromatic), 5.43 (dd, 1H,
© 2004 NRC Canada

Bérces et al.
1169
77.2 (C-3), 74.3 (C-5), 73.8 (C-4), 71.6 (C-2), 64.1 (C-6),
27.1, 26.0 (2O2C(CH3)2), 23.7 (CH3CH2S), 19.9, 19.8
(2ArCH3), 14.9 (CH3CH2S). FAB-MS calcd. for C29H36O7S:
528.2181; found m/z: 529.1 (m + H+), 467.1 (M – Set+) 133
(DiMeBz+). HR-MS calcd. for C29H36O7SNa: 551.2079;
found: 551.2065.
purified by preparative TLC eluting with hexanes : ethyl ac-
etate (80:20) to yield a viscous oil 19 (12 mg, 22% assuming
100% yields for all polymer steps including linker attach-
1
ment). [α]D –75.0 (c, 0.06, CHCl3). H NMR (ppm) δ: 7.27
(qAB, 4H, J = 8.1 Hz, ArH-DOX), 7.22 (t, 1H, J = 7.3 Hz,
J = 8.0 Hz, Bzp), 7.18 (t, 1H, J = 7.3, J = 8.0 Hz, Bzp), 7.06
(d, 2H, J = 8.0 Hz, Bzm), 7.01 (d, 2H, J = 7.3 Hz, Bzm), 5.39
(brt, 1H, H-2), 5.08 (s, 2H, CH2C6H4-DOX), 4.94 (d, 1H,
J = 11.7 Hz, CHHC6H4-DOX), 4.67 (m, 2H, H-6ab), 4.57
MPEGDOXyl 2,6-di-O-(2,6-dimethylbenzoyl)-
3,4-O-isopropylidene-␤-D-galactopyranoside
(18)
(d, 1H, J = 11.7 Hz, CHHC6H4-DOX), 4.52 (d, 1H, J1,2
8.8 Hz, H-1), 4.28 (m, 2H, H-3, H-4), 4.15 (t, 1H, J4,5+5,6
=
=
10.2 Hz, H-5), 2.36, 2.26 (2s, 12H, ArCH3), 2.11 (s, 3H,
COCH3), 1.66, 1.38 (2s, 6H, O2C(CH3)2). 13C NMR (ppm)
δ: 170.8 (COCH3), 169.6, 168.4 (2COAr), 136.7, 135.5
(2CH2-CAr-DOX), 135.0, 135.1 (2Bzip), 133.6, 133.3
(2Bzo), 129.6, 129.3 (2Bzp), 128.2, 128.1 (CHAr-DOX),
127.6, 127.4 (2Bzm), 110.9 (O2C(CH3)2), 99.0 (C-1), 77.2
(C-3), 73.8 (C-4), 73.0 (C-2), 71.1 (C-5), 70.1 (CH2C6H4),
Method 1
Donor 17 (77 mg, 0.15 mmol), 1 (512 mg, 0.1 mmol), and
activated 4 Å molecular sieves (about 300 mg) were dried
together overnight at high vacuum. Dichloromethane (3 mL)
was added and the mixture cooled in an ice–salt bath. After
stirring for 0.5 h, NIS (112 mg, 0.5 mmol) was added fol-
lowed by TfOH (12 µL, 0.14 mmol). After stirring for
30 min, the reaction was quenched with DIPEA (50 µL) and
the polymer precipitated with TBME (40 mL). The solid
was recovered by filtration and after rinsing with TBME was
reprecipitated from absolute ethanol (25 mL). The resulting
solid was collected by filtration and after washing with cold
ethanol and diethyl ether was taken up in CH2Cl2, filtered,
66.0
(CH2C6H4-DOX),
64.1
(C-6),
27.7,
26.6,
(2O2C(CH3)2), 21.0 (COCH3), 19.8, 19.6 (2ArCH3). FAB-
MS: +ve 669 (M + Na+), 467 (M – DOX+), 133 (DiMeBz+).
HR-MS calcd. for C37H42O10Na: 669.2676; found: 669.2715.
Ethyl 2,6-di-O-(2,6-dimethoxylbenzoyl)-3,4-
O-isopropylidene-1-thio-␤-D-galactopyrano-
side (21)
1
and evaporated to yield 18 (490 mg, 88%). Partial H NMR
(ppm) δ: 7.31 (m, 4H, ArH-DOX), 7.24 (t, 1H, J = 7.8 Hz,
Bzp), 7.20 (t, 1H, J = 7.8 Hz, Bzp), 7.08 (d, 2H, J = 7.8 Hz,
Bzm), 7.03 (d, 2H, J = 7.8 Hz, Bzm), 5.37 (brt, 1H, H-2),
4.95 (d, 1H, J = 11.7 Hz, CHHC6H4-DOX), 4.70 (m, 1H, H-
6a), 4.65 (m, 1H, H-6b), 4.58 (d, 1H, J = 11.7 Hz,
CHHC6H4-DOX), 4.56 (s, 2H, CH2C6H4-DOX), 4.54 (d, 1H,
J1,2 = 7.8 Hz, H-1), 4.29 (m, 2H, H-3, H-4), 4.17 (t, 1H,
J4,5+5,6 = 9.8 Hz, H-5), 2.39, 2.29 (2s, 12H, ArCH3), 1.68,
1.40 (2s, 6H, O2C(CH3)2).
Diester 21 was made from diol 16 as for 17 (750 mg,
1
25%). [α]D = 18.2 (c 0.2, CHCl3). H NMR (ppm) δ: 7.27
(m, 2H, Bz), 6.53 (m, 4H, Bz), 5.34 (brt, 1H, J = 6 Hz, H-2),
4.73 (d, 1H, J1,2 = 7.0 Hz, H-1), 4.57 (m, 2H, H-6ab), 4.41
(brt, 1H, J = 5.7 Hz, H-3), 4.31 (brd, 1H, J3,4 = 4.5 Hz, H-
4), 4.14 (t, 1H, J4,5+5,6 = 12.2 Hz, H-5), 3.80, 3.76 (2s, 12H,
ArOCH3), 2.76 (m, 2H, CH3CH2S), 1.61, 1.39 (2s, 6H,
O2C(CH3)2), 1.27 (t, 3H, J = 7.3 Hz, CH3CH2S). 13C NMR
(ppm) δ:, 166.2, 164.7 (2COAr), 157.6, 157.2 (2Bzo), 131.3,
131.0 (2Bzp), 112.8, 112.2 (2Bzip), 110.5 (O2C(CH3)2),
103.9, 103.8 (2Bzm), 82.6 (C-1), 75.6 (C-3), 73.2 (C-5), 72.8
(C-4), 72.3 (C-2), 64.0 (C-6), 56.0, 55.9 (2ArOCH3), 27.8,
26.6 (2O2C(CH3)2), 23.7 (CH3CH2S), 14.8 (CH3CH2S).
FAB- MS calcd. for C29H36O11S: 592; found m/z: 615.2
(M + Na+), 531.2 (M – SEt+), 165 (DiMeOBz+). HR-MS
calcd. for C29H36O11S: 592.2115; found: 592.2128.
Method 2
Orthoester 20 was treated as for 18 above except no NIS,
extra donor, or 1 were added.
6-O-(2,6-Dimethylbenzoyl)-3,4-O-isopropyli-
dene-␤-D-galactopyranosyl-1,2-O-MPEGDOXyl-
ortho-(2,6-dimethylbenzoate) (20)
Orthoester 20 was prepared as for 18 except 0.7 equiv. of
TfOH was used. Partial 1H NMR (ppm) δ: 5.80 (d, 1H, J1,2
=
MPEGDOXyl 2,6-di-O-(2,6-dimethoxylben-
zoyl-3,4-O-isopropylidene-␤-D-galactopy-
ranoside (22)
4.9 Hz, H-1), 4.38 (m, 1H, H-2), 1.68, 1.40 (2s, 6H,
O2C(CH3)2).
Donor 21 was reacted with 1 as for 18 to yield 22 and 23.
22: partial H NMR (ppm) δ: 7.32 (m, 2H, Bz), 7.30 (d, 2H,
J = 8.1 Hz, ArH-DOX), 7.22 (d, 2H, J = 8.1 Hz, ArH-DOX),
6.54 (m, 4H, Bz), 5.27 (brt, 1H, J = 6.1 Hz, H-2), 4.93 (d,
1H, J = 12.0 Hz, CHH-DOX), 4.38 (brt, 1H, H-3), 4.30 (dd,
1H, J3,4 = 5.8 Hz, J4,5 = 1.9 Hz, H-4), 4.11 (brt, 1H, J =
4.4 Hz, H-5), 1.60, 1.36 (2s, 6H, O2C(CH3)2).
[4-O-Acetoxymethyl]benzyl 2,6-di-O-(2,6-
dimethylbenzoyl)-3,4-O-isopropylidene(-␤-D-
galactopyranoside (19)
1
Compound 18 (480 mg, 0.087 mmol) was dissolved in
CH2Cl2 (2.0 mL) and Ac2O (2.0 mL) under an atmosphere
of argon at RT. To this solution was added Sc(OTf)3 (25 mg,
0.02 mmol) and the stirring continued for 4 h. After cooling
with an ice bath, the polymer was precipitated with TBME
(60 mL) and collected by filtration. The filtrate was evapo-
rated. The solid was reprecipitated from ethanol (40 mL)
and collected by filtration. The filtrate was combined with
the residue from TBME and evaporated. The residue was
MPEGDOXyl 3,4-O-isopropylidene-␤-D-
galactopyranoside (24)
Diester 22 (98 mg, 0.017 mmol) was dissolved in 1 mol/L
LiOH in methanol (2 mL) and heated at 50 °C for 16 h un-
© 2004 NRC Canada

1170
Can. J. Chem. Vol. 82, 2004
(c) A.F. Bochkov, V.I. Betaneli, N.K. Kochetkov, and N.K.
Izvestiya. Akad. Nauk SSSR Ser. Khim. 6, 1299 (1974) [In
English] ; [In Russian, p. 1379]; (d) J. Madaj, A. Trynda, M.
Jankowska, and A. WiÑniewski. Carbohydr. Res. 337, 1495
(2002).
der an atmosphere of argon. The heat was removed and the
mixture cooled in an ice bath. The polymer was recovered
by precipitation with TBME and filtration followed by dis-
solution in warm absolute ethanol, filtration while warm,
and cooling. The resulting solid was collected by filtration
and after rinsing with ethanol and diethyl ether was dis-
solved in CH2Cl2, filtered, and evaporated to yield 24
12. (a) M.Z. Liu, H.N. Fan, Z.W. Guo, and Y.Z. Hui. Carbohydr.
Res. 290, 233 (1996); (b) R.T. Lee and Y.C. Lee. Carbohydr.
Res. 271, 131 (1995).
13. S. Deng, B. Yu, J. Xie, and Y. Hui. J. Org. Chem. 64, 7265
(1999).
14. (a) R.T. Brown, N.E. Carter, S.P. Mayalarp, and F. Scheinmann.
Tetrahedron, 56, 7591 (2000); (b) J.R. Ferguson, J.R. Harding,
D.A. Killick, K.W. Lumbard, F. Scheinmann, and A.V.
Stachulski. J. Chem. Soc. Perkin Trans. 1, 3037 (2001).
15. (a) H. Kunz and A. Harreus. Liebigs Ann. Chem. 41 (1982);
(b) S. Sato, T. Nunomura, Y. Ito, and T. Ogawa. Tetrahedron
Lett. 29, 4097 (1988); (c) S. Sato, Y. Ito, and T. Ogawa. Tetra-
hedron Lett. 29, 5267 (1988); (d) S. Nunomura and T. Ogawa.
Tetrahedron Lett. 29, 5681 (1988); (e) M.J. Thompson, E.J.
Hutchinson, T.H. Stratford, W.B. Bowler, and G.M. Blackburn.
Tetrahedron Lett. 45, 1207 (2004).
1
(90 mg, 97%). Partial H NMR (ppm) δ: 7.32 (s, 4H, ArH-
DOX), 4.89 (d, 1H, J = 11.6 Hz, CHH-DOX), 4.25 (brt, 1H,
H-3), 4.14 (dd, 1H, J3,4 = 5.6 Hz, J4,5 = 1.9 Hz, H-4), 4.06
(brt, 1H, J = 5.9 Hz, H-5), 3.97 (m, 1H, H-6a), 1.52, 1.34
(2s, 6H, O2C(CH3)2).
Acknowledgements
The authors gratefully acknowledge the use of the
VPP770 Fujitsu parallel computer situated in the RIKEN
Computer Center. This work was partly supported by the
PENCE program Canada.
16. P. Zimmermann, R. Bommer, T. Bar, and R.R. Schmidt. J.
References
Carbohydr. Chem. 7, 435 (1988).
1. N.K. Kochetkov. In Studies in natural products chemistry.
Vol. 14. Edited by. A. Rahman. Elsevier, Amsterdam. 1994.
pp. 201–266.
17. (a) S. Nunomura, M. Mori, Y. Ito, and T. Ogawa. Tetrahedron
Lett. 30, 5619 (1989); (b) S. Nunomura, M. Mori, Y. Ito, and
T. Ogawa. Tetrahedron Lett. 30, 6713 (1989).
2. (a) H. Paulsen and E. Schüttpelz. Chem. Ber. 112, 3214
(1979); (b) D. Crich, Z. Dai, and S. Gastaldi. J. Org. Chem.
64, 5224 (1999).
18. P.J. Garegg and I. Kvarnstorm. Acta Chem. Scand. Ser. B,
B30, 655 (1976).
19. (a) N.I. Uvarova, G.I. Oshitok, and G.B. Elyakov. Carbohydr.
Res. 27, 79 (1973); (b) T.K. Lindhorst. J. Carbohydr. Chem.
16, 237 (1997); (c) A.F. Bochkov, V.I. Betaneli, and N.K.
Kochetkov. Bioorgan. Khim. 3, 39 (1977). [In English].
20. (a) J. Banoub and D.R. Bundle. Can. J. Chem. 57, 2091 (1979);
(b) V. Magnus, D. Vikiƒ-Topiƒ, S. Iskriƒ, and S. Kveder.
Carbohydr. Res. 114, 209 (1983); (c) Z. Yang, W. Lin, and B.
Yu. Carbohydr. Res. 329, 879 (2000).
3. T. Nukada, A. Bérces, M.Z. Zgierski, and D.M. Whitfield. J.
Am. Chem. Soc. 120, 13 291 (1998).
4. D.M. Whitfield and S.P. Douglas. Glycoconjugate J. 13, 5
(1996).
5. R.U. Lemieux. Chem. Can. 16, 14 (1964).
6. (a) G.H. Veeneman. In Carbohydrate chemistry. Edited by
G.-J. Boons. Blackie Academic and Professional, London.
1998. pp. 98–174; (b) J.-C. Jacquinet and P. Sinaÿ. J. Chem.
Soc. Perkin Trans. 1, 314 (1979).
21. P.H. Amvam-Zollo and P. Sinaÿ. Carbohydr. Res. 150, 199
(1986).
7. (a) F.J. Urban, B.S. Moore, and R. Breitenbach. Tetrahedron
Lett. 31, 4421 (1990); (b) A.Ya. Khorlin, V.A. Nesmeyanov,
and S.E. Zurabyan. Carbohydr. Res. 43, 69 (1975); (c) V.
KÍen, A. Fišerová, C. Augé, P. Sedmera, V. Havlíꢀek, and P.
Šíma. Bioorg. Med. Chem. Lett. 4, 869 (1996); (d) T. Ziegler,
P. Kovac, and C.P.J. Glaudemans. J. Carbohydr. Chem. 9, 135
(1990); (e) N.E. Nifant’ev, E.A. Khatuntseva, A.S. Shahkov,
and K. Bock. Carbohydr. Lett. 1, 399 (1990); (f) U. Nilsson,
A.K. Ray, and G. Magnusson. Carbohydr. Res. 252, 117
(1994); (g) V. Pozsgay, J.-R. Brisson, and H.J. Jennings. Can.
J. Chem. 65, 2764 (1987); (h) N.I. Uvarova, G.I. Oshitok, and
G.B. Elyakov. Carbohydr. Res. 27, 79 (1973).
22. P.J. Garegg and I. Kvarnström. Acta Chem. Scand. Ser. B,
B31, 509 (1977).
23. A. Bérces, T. Nukada, and D.M. Whitfield. J. Am. Chem. Soc.
123, 5460 (2001).
24. (a) H. Paulsen. Angew. Chem. Int. Ed. Engl. 21, 155 (1982);
(b) T. Ishikawa and H.G. Fletcher. J. Org. Chem. 34, 563
(1969).
25. (a) S. Morpurgo, M. Bossa, and G.O. Porpurgo. Phys. Chem.
Chem. Phys. 3, 4898 (2001); (b) S. Morpurgo, M. Brahimi, M.
Bossa, and G.O. Morpurgo. Phys. Chem. Chem. Phys. 2, 2707
(2000).
26. A.F. Bochkov, V.I. Betaneli, and N.K. Kochetkov. Bioorgan.
8. D.M. Whitfield, S.P. Douglas, and J.J. Krepinsky. 17th Interna-
tional Carbohydrate Symposium, Ottawa, Ont. 1994.
9. (a) J.J. Krepinsky and S.P. Douglas. In Solid support oligo-
saccharide synthesis and combinatorial libraries. Edited by
P.H. Seeberger. Wiley-Interscience, New York. 2001. pp. 175–
199; (b) J.J. Krepinsky and S.P. Douglas. In Carbohydrates in
chemistry and biology. Vol. 1. Edited by B. Ernst, G.W. Hart,
and P. Sinaÿ. Wiley-VCH, Weinheim. 2000. pp. 239–265.
10. T. Nukada, A. Bérces, and D.M. Whitfield. J. Org. Chem. 64,
9030 (1999).
11. (a) D.M. Whitfield, S.P. Douglas, T.H. Tang, I.G. Csizmadia,
H.Y.S. Pang, F.L. Moolten, and J.J. Krepinsky. Can. J. Chem.
72, 2225 (1994); (b) N.M. Spijker, J.E.M. Basten, and C.A.A.
van Boeckel. Recl. Trav. Chim. Pays–Bas, 112, 611 (1993);
Khim. 2, 671 (1976). [In English].
27. E. Eichler, F. Yan, J. Sealy, and D.M. Whitfield. Tetrahedron,
57, 6679 (2001).
28. S. Mehta and D.M. Whitfield. Tetrahedron Lett. 39, 5907 (1998).
29. D. Crich, Z. Dai, and S. Gastaldi. J. Org. Chem. 64, 5224
(1999).
30. (a) J.E. Wallace and L.R. Schroeder. J. Chem. Soc. Perkin
Trans. 2, 795 (1977); (b) L.G. Green and S.V. Ley. In Carbo-
hydrates in chemistry and biology. Vol. 1. Edited by B. Ernst,
G.W. Hart, and P. Sinaÿ. Wiley-VCH, Weinheim. 2000.
pp. 427–448.
31. R.U. Lemieux and A.R. Morgan. Can. J. Chem. 43, 2199
(1965).
32. (a) H.S. Isbell and H.L. Frush. J. Res. Natl. Bur. Stand. 43,
© 2004 NRC Canada

Bérces et al.
161 (1949); (b) Y. Zeng, J. Ning, and F. Kong. Tetrahedron
1171
37. J.J. Krepinsky, S.P. Douglas, and D.M. Whitfield. Methods
Enzymol. 242, 281 (1994).
Lett. 43, 3729 (2002).
33. T. Nukada, A. Bérces, and D.M. Whitfield. Carbohydr. Res.
38. G. Hodosi and J.J. Krepinsky. Synlett, 159 (1996).
39. R.R. Schmidt and M. Stumpp. Liebigs Ann. Chem. 1249
(1983).
40. Y. Buskas and P. Konradsson. J. Carbohydr. Chem. 19, 25
(2000).
337, 765 (2002).
34. J. March. Advanced organic chemistry. 2nd ed. McGraw Hill,
London. 1977. p. 248.
35. T. Nukada, A. Bérces, and D.M. Whitfield. Tetrahedron, 57,
477 (2001).
36. P. Pulay, G. Fogarasi, F. Pang, and J.E. Boggs. J. Am. Chem.
Soc. 101, 2550 (1979).
© 2004 NRC Canada